COMM. MATH. SCI.

c 2003 International Press

Vol. 1, No. 1, pp. 6873
ON THE EQUATION SATISFIED BY A STEADY
PRANDTL-MUNK VORTEX SHEET

MILTON C. LOPES FILHO

, HELENA J. NUSSENZVEIG LOPES

, AND MAX O. SOUZA

Abstract. We show that the vorticity distribution obtained by minimizing the induced drag on
a wing, the so called Prandtl-Munk vortex sheet, is not a travelling-wave weak solution of the Euler
equations, contrary to what has been claimed by a number of authors. Instead, it is a weak solution
of a non-homogeneous Euler equation, where the forcing term represents a tension force applied
to the tips. This is consistent with a heuristic argument due to Saman. Thus, the notion of weak
solution captures the correct physical behavior in this case.
The Prandtl-Munk vortex sheet is notorious in uid dynamics as the one that
generates a circulation distribution that minimizes the induced drag on a plane wing
[1, 8]. The vorticity associated to the Prandtl-Munk vortex sheet is given by

0
(x
1
, x
2
) =
x
1
_
1 x
2
1

(1,1)
(x
1
)
0
(x
2
), (1)
where
(1,1)
represents the characteristic function of the interval (1, 1) and
0
is
the Dirac delta at x
2
= 0.
The associated velocity, given by the Biot-Savart law, can be evaluated at points
on the sheet by means of a principal value integral and is found to be constant and
equal to (0, 1/2) [8]. Hence a natural candidate for a weak solution is determined
by the steadily translating vorticity prole:
(x
1
, x
2
, t) =
0
(x
1
, x
2
+
t
2
). (2)
A number of authors [3, 6] have claimed, albeit with some doubts, that (2) is indeed
a weak solution of the steady Euler equations. However, the numerical experiments
performed by Krasny [3] suggest that the (physically correct) solution of the Euler
equation with (1) as initial data is not steady, actually rolling-up at the tips for positive
time. Later, Saman [8] gave a heuristic argument, based on vortex principles, as to
why the steady solution (2) is not dynamically consistent. This scenario suggests
that the Prandtl-Munk vortex prole (1) is an example of nonuniqueness for the weak
formulation of vortex sheet evolution. We shall show that (2) is actually a weak
solution of a nonhomogeneous Euler equation, with forcing term consistent with that
obtained by Saman, thus ruling out the Prandtl-Munk vortex sheet as an example
of nonuniqueness.
It is not known to which degree unphysical ows may satisfy the standard notion
of weak solution to the Euler equations. Examples of (unphysical) weak solutions
with kinetic energy growing in time have been obtained [9, 11] in a situation con-
siderably more irregular than vortex sheet dynamics. The main point of the present

Received: April 15, 2002; Accepted (in revised version): April 25, 2002.

Departamento de Matematica, IMECC-UNICAMP. Caixa Postal 6065, Campinas, SP 13081-970,

Brasil (mlopes@ime.unicamp.br).

Departamento de Matematica, IMECC-UNICAMP. Caixa Postal 6065, Campinas, SP 13081-970,

Brasil (hlopes@ime.unicamp.br).

Departamento de Matematica, PUC-Rio, Rua Marques de Sao Vicente, 225 Rio de Janeiro -
RJ, 22453-900, Brasil (msouza@mat.puc-rio.br). Present address: Depto. de Matematica Aplicada,
UFF, Rua Mario Santos Braga s/n, Niteroi, RJ 24020-140, Brasil.
68
M. C. LOPES FILHO, H. J. NUSSENZVEIG LOPES, AND M. O. SOUZA 69
paper is the fact that the notion of weak solution is discriminating enough to distin-
guish the physically correct evolution for the Prandtl-Munk vortex sheet, something
which cannot be done through the Birkho-Rott model alone. A dierent example
of nonphysical weak solution in the context of vortex sheets has been proposed and
numerically studied, see Lopes Filho et al. [4].
The standard denition of a weak solution of the Euler equations uses velocity
as the main dynamic variable [5]. We observe that it is not easy to check, using this
notion of weak solution, whether (2) is or is not a weak solution of the Euler equations.
Indeed, despite the fact that the vorticity associated to the Prandtl-Munk vortex sheet
has the particularly simple form (1), the associated velocity is fairly complicated [8].
For this reason, we choose to work with the vorticity equation, and we will use the
weak vorticity formulation originally introduced by Delort [2] and reformulated by
Schochet [10] stated as follows: a measure L

([(0, ); B/(R
2
) H
1
loc
(R
2
))
is a weak solution of the inhomogeneous vorticity formulation of the incompressible
2D Euler equations with initial data
0
B/
c
(R
2
) H
1
(R
2
), and forcing F
L

([(0, ); W
1,1
loc
(R
2
)) if, for any test function C

c
([(0, ) R
2
), we have:
_

0
__
R
2

t
d(x, t) +
_
R
2
_
R
2
H

d(x, t) d(y, t)
_
dt+
_
R
2
(x, 0)d
0
(x) =
_

0
_
R
2

dF(x, t)dt, (3)

where
H

(x, y, t)
(x, t) (y, t)
4[x y[

(x y)

[x y[
.
In the homogeneous case, this formulation was shown to be equivalent to the standard
velocity one, as long as the velocity is (L
2
loc
(R
2
))
2
, by Delort [2]. However, the proof
carries over in a straightforward manner to the inhomogeneous case.
Using the formulation above, we shall show that the velocity associated to (2)
solves, in a weak sense, the system of equations:
_
u
t
+ u u = p + F
div u = 0,
with the forcing term
F(x
1
, x
2
) = /8
_
_

0
(x
1
1, x
2
+ t/2)
0
(x
1
+ 1, x
2
+ t/2)
0
_
_
. (4)
The factor in (4) should be compared with equation (7) in 6.3 of Saman [8],
keeping in mind that, in our case, we have U = 1/2 and a = 1. Thus, in accordance
with Samans analysis, the forcing given by (4) can be thought of as a tension applied
near the tips of the vortex sheet.
We begin by choosing a test function of the form
(x
1
, x
2
, t) = (t)
_
x
1
, x
2
+
t
2
_
,
where and are C

c
in [0, ) and R
2
, respectively. We notice that nite sums
of test functions having this form are dense in the set of all C

c
([0, ) R
2
) test
functions.
We now compute the right-hand side of (3) with given by (2):
70 ON THE EQUATION SATISFIED BY A STEADY PRANDTL-MUNK VORTEX SHEET
(i)
_

0
_
R
2

t
d(x, t)dt =
_

0
_
1
1
_

(t)(x
1
, 0) + (t)
1
2

x
2
(x
1
, 0)
_
x
1
_
1 x
2
1
dx
1
dt =

_
1
1
(0)(x
1
, 0)
x
1
_
1 x
2
1
dx
1
+
_

0
(t)
1
2
_
1
1

x
2
(x
1
, 0)
x
1
_
1 x
2
1
dx
1
dt;
(ii)
_

0
_
R
2
_
R
2
H

d(x, t) d(y, t)dt =

_

0
(t)
1
4
_
1
1
_
1
1

x
2
(x
1
, 0)
x
2
(y
1
, 0)
x
1
y
1
x
1
_
1 x
2
1
y
1
_
1 y
2
1
dx
1
dy
1
dt;
(iii)
_
R
2
(x, 0)d
0
(x) =
_
1
1
(0)(x
1
, 0)
x
1
_
1 x
2
1
dx
1
.
adding the three terms, dropping the subscript in x
1
and in y
1
, and writing
x
2
= (x),
we obtain:
_

0
__
R
2

t
d(x, t) +
_
R
2
_
R
2
H

d(x, t) d(y, t)
_
dt +
_
R
2
(x, 0)d
0
(x) =
1
2
_
1
1
(x)
x

1 x
2
dx+
1
4
_
1
1
_
1
1
(x) (y)
x y
x

1 x
2
y
_
1 y
2
dxdy = S, ). (5)
We now show that
S =

8
(
1

1
),
We start by observing that the support of S is obviously contained in the closed
interval [1, 1].
Next we observe that S is an odd distribution, that is, if is an even test function
then S, ) = 0. Indeed, the rst integral vanishes trivially, whereas to see that the
second vanishes it suces to use the change of variables (x, y) (x, y).
Now we show that S, ) vanishes for any which is a polynomial vanishing at
x = 1 and x = 1 simultaneously. Clearly it is enough to verify this for all test
functions of the form
n
(x) = (1 x
2
)x
2n1
, n = 1, 2, . . ..
Observe that, for all k = 1, 2, . . ., we have that
_
1
1
x
2k

1 x
2
dx =
_

0
(cos )
2k
d =
(2k 1)!!
(2k)!!
,
M. C. LOPES FILHO, H. J. NUSSENZVEIG LOPES, AND M. O. SOUZA 71
where
(2n)!! =
n

=1
(2), (2n 1)!! =
n

=1
(2 1)
and the second equality is a well-known calculus identity [7].
Hence, the rst term in (5) can be evaluated as
_
1
1

n
(x)
x

1 x
2
dx =
_
(2n 1)!!
(2n)!!

(2n + 1)!!
(2n + 2)!!
_
.
Since

n
(x)
n
(y)
x y
=
2n2

=0
x

y
2n2

2n

=0
x

y
2n
;
the double integral in (5) can be evaluated, after noticing that even powers integrate
to zero, as follows:
_
1
1
_
1
1

n
(x)
n
(y)
x y
x

1 x
2
y
_
1 y
2
dxdy =
n1

=1
_
1
1
_
1
1
x
21
y
2n12
x

1 x
2
y
_
1 y
2
dxdy+

=1
_
1
1
_
1
1
x
21
y
2n+12
x

1 x
2
y
_
1 y
2
dxdy =

2
_
n1

=1
(2 1)!!
(2)!!
(2n 2 1)!!
(2n 2)!!

n

=1
(2 1)!!
(2)!!
(2n 2 + 1)!!
(2n 2 + 2)!!
_
.
Thus, we want to show that
0 = S,
n
) =
1
2

_
(2n 1)!!
(2n)!!

(2n + 1)!!
(2n + 2)!!
_
+
1
4

2
_
n1

=1
(2 1)!!
(2)!!
(2n 2 1)!!
(2n 2)!!

n

=1
(2 1)!!
(2)!!
(2n 2 + 1)!!
(2n 2 + 2)!!
_
. (6)
Set
a
n
=
(2n + 1)!!
(2n + 2)!!
+
1
2
n

=1
(2n + 1 2)!!
(2n + 2 2)!!
(2 1)!!
(2)!!
,
for n 2, and set a
0
= a
1
= 1/2. With these denitions, (6) is equivalent to
a
n1
a
n
= 0.
Indeed, we now show that, for all n = 2, 3, . . ., a
n
= 1/2.
72 ON THE EQUATION SATISFIED BY A STEADY PRANDTL-MUNK VORTEX SHEET
Write:
b
n
=
(2n + 1)!!
(2n + 2)!!
.
Then, with this notation we have:
a
n
= b
n
+
1
2
n

=1
b
n
b
1
,
so that what we need to show is that the right-hand side of the equality above is
identically equal to 1/2, for n = 2, 3, . . .. We will use a generating function argument.
We introduce the function:
f(x) =
1
x
_
1

1 x
1
_
.
It can be easily checked that the Taylor expansion of f is
f(x) =

n=0
b
n
x
n
.
Therefore, the identity sought reduces, through term-by-term identication of the
Taylor coecients, to the following algebraic identity:
f(x)
1
2

3x
8
+
x
2
((f(x))
2

1
4
) =
1
2
_
1
1 x
1 x
_
.
Since the
n
are dense in C

c
((1, 1)), S vanishes against any such test functions.
Thus, the support of S is contained in 1, 1 and, in this case, S must be of the
form
S =
N

k=0

k
(
(k)
1
(1)
k

(k)
1
),
where
k
R, and we have taken advantage of the fact that S is odd. Applying S to
p(x) = x(1 x
2
)
n
, we see that
n
= 0, n 1. Finally, taking (x) = x shows that

0
= /8.
Reverting back to our original notation, we have that (3) yields
_

0
(t)
_

x
2
(1, 0)

8

x
2
(1, 0)
_
dt =
_

0
_
R
2

dF(x, t)dt,
as claimed.
Acknowledgements. The authors would like to thank Robert Krasny and Oscar
Orellana for their helpful comments. Part of this work was done during a visit of
the third author to the rst two, who thanks UNICAMP for the hospitality during
the visit. M.C.L.F.s research is supported in part by CNPq grant # 300.962/91-6.
H.J.N.L.s research is supported in part by CNPq grant # 300.158/93-9. This work
was also supported by FAPESP grant # 97/13855-0 and by PRONEX.
M. C. LOPES FILHO, H. J. NUSSENZVEIG LOPES, AND M. O. SOUZA 73
REFERENCES
[1] G.K. Batchelor, An Introduction to Fluid Dynamics. Cambridge University Press, 1967.
[2] J.-M. Delort, Existence de nappes de tourbillon en dimension deux. J. of Amer. Math. Soc.,
4:553586, 1991.
[3] R. Krasny, Computation of vortex sheet roll-up in Tretz plane. J. Fluid Mech., 184:123155,
1987.
[4] M.C. Lopes Filho, J. Lowengrub, H.J. Nussenzveig Lopes, and Y. Zheng, Numerical evidence
for the nonunique evolution of vortex sheets in the plane. Relatorio de Pesquisa 44/99,
IMECC-UNICAMP, 1999.
[5] C. Marchioro and M. Pulvirenti, Mathematical Theory of Incompressible Nonviscous Fluids.
Springer-Verlag, 1994.
[6] R.E. Meyer, Introduction to Mathematical Fluid Dynamics. Dover, 1982.
[7] I.M. Ryzhik and I.S. Gradshteyn, Table of Integrals, Series and Products. Academic Press,
1994.
[8] P.G. Saman, Vortex Dynamics. Cambridge University Press, 1992.
[9] V. Scheer, An inviscid ow with compact support in space-time. J. of Geom. Anal., 3:343401,
1993.
[10] S. Schochet, The weak vorticity formulation of the 2d Euler equations and concentration-
cancellation. Comm P.D.E., 20:10771104, 1995.
[11] A. Shnirelman, On the nonuniqueness of weak solutions of the Euler equations. Comm. Pure
Appl. Math., 50:12611286, 1997.